Nwāchā Munā: A Devanagari Speech Corpus and Proximal Transfer Benchmark for Nepal Bhasha ASR
This work addresses the severe scarcity of annotated speech resources for Nepal Bhasha, an endangered language, providing a dataset and a computationally efficient ASR solution for the Newari community.
This paper introduces Nwāchā Munā, a 5.39-hour manually transcribed Devanagari speech corpus for Nepal Bhasha, and establishes a benchmark for ASR. They show that fine-tuning a Nepali Conformer model reduces the Character Error Rate (CER) from 52.54% to 17.59%, matching the performance of the multilingual Whisper-Small model.
Nepal Bhasha (Newari), an endangered language of the Kathmandu Valley, remains digitally marginalized due to the severe scarcity of annotated speech resources. In this work, we introduce Nwāchā Munā, a newly curated 5.39-hour manually transcribed Devanagari speech corpus for Nepal Bhasha, and establish the first benchmark using script-preserving acoustic modeling. We investigate whether proximal cross-lingual transfer from a geographically and linguistically adjacent language (Nepali) can rival large-scale multilingual pretraining in an ultra-low-resource Automatic Speech Recognition (ASR) setting. Fine-tuning a Nepali Conformer model reduces the Character Error Rate (CER) from a 52.54% zero-shot baseline to 17.59% with data augmentation, effectively matching the performance of the multilingual Whisper-Small model despite utilizing significantly fewer parameters. Our findings demonstrate that proximal transfer within South Asian language clusters serves as a computationally efficient alternative to massive multilingual models. We openly release the dataset and benchmarks to digitally enable the Newari community and foster further research in Nepal Bhasha.